I. Core Research Summary

• Revealing the underlying formation mechanism of corner defects

On curved paths, the outer trajectory is longer and deposits more material, leading to protrusions, while the inner trajectory is shorter and lacks sufficient material, resulting in voids. In multi-layer printing, defects accumulate layer by layer; the magnification effect is most pronounced for small sharp angles (15°), whereas for large bends of 90° or greater, the impact of material imbalance is negligible. Platform acceleration and deceleration have no significant effect on the deposited volume of structures larger than 100 μm.

• Quantify the minimum chamfer standards for each angle and provide engineering‑grade parameters.

(1) 15° Ultra-Sharp Angle: A simple chamfer cannot eliminate the defect; it must be split into an independent single-segment toolpath.

(2) 30° angle: Rounded corners ≥ line width can completely suppress overflow and voids.

(3) 45° angle: Uniform deposition can be achieved when the fillet radius is at least half the line width.

(4) Large turning angles of 90°/135°: No additional fillets are required, and deposition is uniform with no noticeable height differences.

• Establish a semi-empirical geometric model for simplified material deposition.

Only the path-length difference is used to calculate the vertical deviation of internal and external deposits, enabling prediction of defect distribution; meanwhile, it is demonstrated that the ink’s self‑leveling effect significantly reduces the model’s theoretical error, representing an advantage unique to aerosol printing compared with extrusion‑based additive manufacturing.

• Three major pathways for optimizing turnkey solutions

By combining layered offset printing trajectories, adding matched fillets at corners, and leaving minute gaps along closed contours, it is possible to stably fabricate star-shaped and irregular three-dimensional micro-walls.

• Multi-ink compatibility verification

Both silver nanoparticle and zinc oxide ceramic inks can be defect-free fabricated using the same process route, making this approach compatible with a wide range of aerosol‑based functional pastes for metals and ceramics.

II. Complete Preparation Process for 3D Microstructures of Aerosols

(1) Printing Procedure

• Nanoparticle ink dilution and formulation: silver/zinc oxide powders are dispersed in a cyclohexanone–terpineol mixed solvent, with ultrasonic treatment employed to eliminate agglomeration.
• Alumina substrate polishing and cleaning, with a 70°C constant-temperature printing workbench;
• AJ300 ultrasonic atomization, with precise sheath‑gas/carrier‑gas ratio control, a 150 μm ceramic nozzle, and a printing speed of 1 mm/s;
• Create the basic outline in CAD, add corresponding fillets based on the included angles, leave a gap at the closure, and switch printing paths layer by layer.
• Micron-scale three-dimensional micro-walls are stacked layer by layer, with each layer being dried and cured at low temperature after printing.
• A white-light interferometer and SEM are used to observe the height and morphology of cross-sectional corners, and to statistically analyze the dimensions of overflow and void defects.

(II) Core Process Advantages of Five‑Atmosphere Aerosol 3D Printing (Compared with FDM Extrusion and Two‑Photon Lithography)

• Support-free freeform micrometer-scale three-dimensional micro-walls

Without the need for sacrificial layers or lithographic masks, a single-step, layer-by-layer stacking approach enables the fabrication of three-dimensional conductive/ceramic structures, thereby simplifying the microdevice fabrication process.

• Controllable formation of micron-scale fine lines

Stably fabricates 20-μm ultrafine walls, suitable for high-precision applications such as EEG electrodes, microsensors, and microfluidics.

• Wide ink compatibility

Metallic silver, oxide ceramics, and conductive polymers can all be printed via atomization, with a single system capable of supporting diverse applications in electronics, biotechnology, and energy storage.

• Low-temperature, gentle molding, compatible with flexible thermosensitive substrates.

Throughout the process, the substrate temperature is maintained at 70°C, preventing high-temperature sintering damage to flexible substrates such as PI and silicone rubber.

• The digitalization path is flexible and adjustable, with low iteration costs.

With CAD, you can instantly modify corner fillets and layered toolpaths—no need to retool—and complete comparative prototyping of multiple designs on the same day.

III. Five Key Technological Innovations

Innovation 1: A Quantitative Benchmark System for Corner Defects

Innovation 2: A Deposition Prediction Model That Also Accounts for the Self-Leveling Effect

Innovation: A Threefold, Flawless Paradigm for Path Optimization

Innovation 4: A Universal Microstructure Manufacturing Approach for Multiple Materials

Innovation 5: Rapid prototyping capability for complex, irregular microstructures

IV. Current Industry Challenges and Corresponding Solutions for Overcoming Them

Challenge 1: In aerosol printing, the distribution of internal and external materials at sharp corners is uneven, leading to severe protrusion and void defects, which result in failure after multiple layers.

Challenge 2: Closing the contour and triggering the shutter at the start and end of each layer leads to localized material buildup, with defects in each layer accumulating and magnifying.

Challenge 3: Lack of standardized guidelines for pathway design, resulting in R&D relying heavily on trial-and-error and consuming significant resources and time.

Challenge 4: Traditional geometric models overestimate defect sizes, resulting in poor prediction accuracy.

Challenge 5: Different metal and ceramic inks require individually optimized process parameters.

V. Future Diversified Industrial Application Scenarios

• High-Density 3D Brain–Machine Microelectrode Array

• Thick electrode for micro lithium-ion batteries

• Microfluidic biochemical sensing chip

• Miniature RF antennas, chip interconnects

• Aerospace Microscale Lightweight Mechanical Metamaterials

• University Micro- and Nanomanufacturing R&D Platform

Figure 1. Aerosol jet (AJ) printing process and examples of defects. (a) Schematic illustration of the AJ printing process employed in this study. An ultrasonic atomizer is used to nebulize a silver nanoparticle ink into an aerosol, which is then conveyed to the deposition head, where the material stream is focused by a sheath gas flow. Subsequently, microdroplets are deposited onto a moving substrate and rapidly dried by heating. (b) CAD schematic of an octagonal star. (c) The inset shows the toolpath contour, corresponding to the two-dimensional cross-section depicted in (b). (d) An AJ‑printed three-dimensional star-shaped geometry generated using the toolpath shown in (c). Regions of the toolpath featuring sharp angles exhibit pronounced protrusions and adjacent voids.

Figure 2. Geometric structure of AJ‑printed two-dimensional lines. (a) Schematic illustration of sharp angles in the planned toolpath, showing overlapping at internal corners and insufficient filling at external corners. Compared with straight‑line cross sections, the overlapping regions are expected to exhibit overfilling, while the void regions may suffer from underfilling. (b) CAD drawing of the experimental setup for AJ‑printed two-dimensional lines. Five different angles—15°, 30°, 45°, 90°, and 135°—were investigated on the same substrate, with corner radii ranging from 0 µm to 1 mm within the toolpath. Each condition was replicated five times. Auxiliary vertical and horizontal reference lines are also included to facilitate alignment during optical surface profilometry. (c) Representative quantitative surface profile data for a single pass at a 30° angle. (d, e) Local magnified views corresponding to the 0 µm corner radius case shown in (c). (f, g) Local magnified views corresponding to the 30 µm corner radius case shown in (c).

Figure 3. Volume and height data for a single‑pass printed line. (a) Schematic illustrating the region of interest (ROI), defined as the path length from point A to point B, with both points equidistant from the projection intersection. These regions are analyzed individually and jointly to compare volume and average height, and the variables used in the mathematical model are shown. (b–d) Show the total volume of AJ‑printed lines within the ROI at bending angles of 15°, 30°, and 45°, respectively, along with the results of linear fits to the data. The linear relationship between volume and ROI length confirms a stable deposition rate, and the platform‑based acceleration/deceleration during the curved segments is not a critical factor in this study. (e–g) Display the fraction of the total volume contributed by the inner‑side AJ‑printed lines within the toolpath at bending angles of 15°, 30°, and 45°; deviations from the 50% mark indicate self‑leveling behavior of the material during drying. (h–j) Present the average height of the printed lines within the ROI at bending angles of 15°, 30°, and 45°, including contributions from both inner and outer volumes. (k–m) Show the height differences between the inner and outer volumes of the toolpath at the same bending angles. All data points in the figures represent the mean and standard deviation of five replicate experiments, with each replicate averaged over pixels within the ROI (the number of pixels ranging from several dozen to several thousand, depending on the ROI length). Similar results for AJ‑printed lines at 90° and 135° bends are provided in Supplementary Figure S2.

Figure 4. Comparison between exponential decay fitting results and predictions from the geometric model. (a) Schematic of the variables used in the mathematical model. (b–d) Experimental data from Figures 3k–m, together with exponential decay fits, for angles of 15°, 30°, and 45°, respectively. (e–g) Predicted height differences at 15°, 30°, and 45°, based on the geometric model assuming no self‑leveling effect during drying. Clearly, the model‑predicted height differences are an order of magnitude larger than the experimentally observed values, underscoring the significance of the self‑leveling effect. All experimental data points in the figure represent the mean and standard deviation of five replicate measurements; for each replicate, the value corresponds to the average pixel intensity within the region of interest (ROI), with the number of pixels ranging from several dozen to several thousand depending on the ROI length. Similar results for AJ print lines at the 90° and 135° bends are provided in Supplementary Figure 2.

Figure 5. High-fidelity three-dimensional structures achieved through a rational AJ‑based toolpath strategy for additive manufacturing. (a–c) Schematic representations of the toolpaths used to print an octagonal star‑shaped 3D model. Note the shutter opening/closing gaps shown in the schematics; these gaps are assigned to different segments of the star in each case (16 segments total). After printing 16 layers, the printing direction is reversed, ultimately completing a unique structure comprising 32 layers. (d–f) Perspective scanning electron microscope images of the octagonal star‑shaped object printed using the toolpaths depicted in (a–c); corresponding top‑view SEM images are provided in Supplementary Figure S4a–S4c. (g–i) Close-up images of selected points along the outer perimeter of the star, as illustrated in (a–c). (j–l) Profile scans obtained via white‑light interferometry of the images in (g–h), revealing the material deposition morphology. (m) Quantitative measurements of the height differences between corner points and the central point were performed for three size conditions: R = 0, R = 20 μm, and R = 30 μm.

Figure 6. A 3D model of Carnegie Mellon University’s mascot, “Scotty the Dog,” printed using different toolpath strategies. (a–b) Schematic diagrams of the toolpaths for Scotty the Dog at corner radii of 0 μm and 20 μm, with selected angular values labeled. Note that for both radius settings, the positions of the shutter‑open and shutter‑close markers—corresponding to gaps in the toolpath—are adjusted to locations where no change in the toolpath occurs; a total of 16 such positions are used. These 16 layers are printed first, after which the same gap positions are reversed in the subsequent 16 layers, resulting in a total of 32 distinct printing layers. (c) SEM oblique‑view image of the CMU “Scotty the Dog” logo printed via aerosol jetting with a toolpath radius of 0 μm; all wall thicknesses are 20 μm. (d) SEM oblique‑view image of the CMU “Scotty the Dog” logo printed via aerosol jetting with a toolpath radius of 20 μm; all wall thicknesses are 20 μm. (e–h) Enlarged views of the region shown in (c). (i–l) Enlarged views of the region shown in (d). 。